Definitions of common named elements in the Graft help center.
A user created workspace for your Machine Learning (ML) project. Contains all of the processing of one or more data sources in Entities to support your business use cases.
A connection to a data set of interest in a use case. Data sources can be internal or public data sets which you have rights to access and process.
Representations of your unstructured data that have been created by applying a machine learning model.
A foundation model, either provided by Graft or via API from a third party. Applying a trunk model to your data generates a series of vectors known as embeddings which can be used for semantic and visual search (depending on the data type) and are the entry point for creating an enrichment to generate predictions
A classifier model which when applied to embeddings can generate a prediction about that data. Graft provides a number of Enrichment models and also supports the creation of custom models by the user
The acquisition of your target data and application of one or more Models
The output of a Enrichment model, generating a classification of the data in the entity, for example, predicting whether an image has a vehicle in it.
The process of taking an unlabelled data set and known characteristics of that data to generate an initial set of predictions which are used as labels to create an Enrichment
A known characteristic of a data point, used to build enrichment models. For example a label field for a clothing catalogue could "Color" and have values (Classes) "Blue", "Green", "Red", "Black", "White" etc.
Goal specific method of quickly creating a use case using Graft selected default settings in a few steps no Machine Learning experience required. For example; creating a visual search app from a datasource containing images. For more information please read Introduction to Apps